Freebayes AF field distrubution.

[Richard Finney]

I have run freebayes SNP caller on a whole genome sample.
It is believed to possibly have cells with genomes that have somatic amplifications or deletions. I am interested in identifying "amplified" chromosomes.

The Freebayes output VCF has the "AF" field : "Estimated allele frequency in the range (0,1]"

Here's a plot of the AF field and counts ...



Y axis is count, X axis is AF for SNPs for several chromosomes.

Why is the distribution bimodal? At 0.5 and at 0.47

Note the minor kinks at 0.33 and 0.67, too.

I'm at a loss. Any thoughts?

[HESmith]

The kinks at 0.33 and 0.67 could represent duplicated chromosome fragments containing the wild-type and variant alleles, respectively. But I have no idea about the bimodal primary peaks. Perhaps 50% is an artifact of a prior expectation (heterozygosity) built into the algorithm?